Biology subjects

Hayes, D. N.

Publications and source records attributed to Hayes, D. N..

7 recordsLinked to original sources

bioRxiv2026

Equitable Health Intelligence: An Open Benchmark of Multi-Population Machine Learning for Omics-Based Cancer Prognosis

PurposeMachine learning (ML) models for omics-based cancer prognosis are often trained on data from predominantly European-ancestry populations, producing biased predictions for other populations and undermining equitable genomic medicine. Existing fairness benchmarks mainly focus on outcome parity rather than predictive performance parity across populations. Public benchmark resources are needed for systematically detecting and mitigating such performance disparities in multi-population cancer prognosis. MethodsWe developed Equitable Health Intelligence (EHI, https://ehiportal.org), an open-source benchmark of multi-population ML for omics-based cancer prognosis. EHI contains 1,475 ML tasks across 40 cancer/pan-cancer types, 4 omics feature sets, 4 clinical endpoints, 5 event-time thresholds, and 3 data-disadvantaged population (DDP) groups relative to a majority European Ancestry population group. Deep neural network models are trained under three multi-population ML schemes (Mixture, Independent, and Transfer Learning), with Naive Transfer included as a no-adaptation control, comprising a total of 10,325 ML experiments. ResultsThe EHI platform provides an interactive environment with visualization and exploratory tools for users to inspect predictive performance disparities between the majority European-ancestry group and data-disadvantaged populations, evaluate the extent to which transfer learning mitigates these disparities, and examine the impact of feature engineering methods across cancer types, omics features, and clinical endpoints. ConclusionEHI is an open, interactive, and extensible benchmark for identifying and addressing performance disparities in multi-population ML for omics-based cancer prognosis. It provides a foundation for a growing ecosystem of methods targeting ML performance disparities arising from biomedical data inequality and population-level distribution shifts, thereby advancing equitable AI in precision oncology.

bioinformatics↗

bioRxiv2025

Lung Adenocarcinoma Just Desserts: An Expanding Pie of Activating Oncogenes or a Layer Cake of Integrated Alterations

The molecular landscape of lung adenocarcinoma (LUAD) is often summarized as a "pie chart" of driver oncogenes, suggesting identification and targeting of oncogenic drivers is the best clinical approach. However, this model oversimplifies LUAD biology. Patients with identical RAS or RAF mutations exhibit heterogeneous signaling influenced by co-mutations, transcriptional programs, and lineage context. We propose a hierarchical framework integrating oncogenes within cellular context by examining canonical EGFR mutations. We defined an EGFR mutation signature (mSig) by identifying differentially expressed genes in EGFR-mutant LUADs. Semi-supervised clustering and machine learning models were used to test performance and reproducibility across independent datasets. We analyzed molecular subtypes, lineage markers, co-occurring mutations, and candidate drug targets in EGFR-mSig positive (+) versus negative (-) tumors. The EGFR mSig showed robust classification performance across datasets (AUROC = 0.83-0.95; mean NPV = 96.3%). Validated unsupervised gene expression subtypes and lung lineage markers were closely aligned with EGFR mSig status. mSig(+) tumors were identified, even in tumors without EGFR mutations. A subset of canonical RAS mutations mirrored the EGFR mutation pattern. EGFR-mutant/mSig(-) tumors were enriched for non-Bronchioid subtypes and had co-mutations in TP53 or RAS. Coordinated mutations were identified including RAS, KEAP1, STK11, TP53, and CDKN2A, supportive of prior reports. In sum, novel EGFR mSig that captures the transcriptional footprint of EGFR activation, revealed a subset of EGFR wildtype LUADs with "mutant-like" features. Lineage-informed classification highlights subtype-dependent oncogene activity and supports new therapeutic strategies. A context-specific association between RAS mutation and expression of the dual-specificity phosphatase gene DUSP4 may have therapeutic potential.

genomics↗

bioRxiv2022

Response to immune checkpoint blockade improved in pre-clinical model of breast cancer after bariatric surgery

Bariatric surgery is becoming more prevalent as a sustainable weight loss approach, with vertical sleeve gastrectomy (VSG) being the first line of surgical intervention. We and others have shown that obesity exacerbates tumor growth while diet-induced weight loss impairs obesity-driven progression. It remains unknown how bariatric surgery-induced weight loss impacts cancer progression or alters responses to therapy. Using a pre-clinical model of diet induced obesity followed by VSG or diet-induced weight loss, breast cancer progression and immune checkpoint blockade therapy was investigated. Weight loss by bariatric surgery or weight matched dietary intervention before tumor engraftment protected against obesity-exacerbated tumor progression. However, VSG was not as effective as dietary intervention in reducing tumor burden despite achieving a similar extent of weight and adiposity loss. Circulating leptin did not associate with changes in tumor burden. Uniquely, tumors in mice that received VSG displayed elevated inflammation and immune checkpoint ligand, PD-L1. Further, mice that received VSG had reduced tumor infiltrating T lymphocytes and cytolysis suggesting an ineffective anti-tumor microenvironment. VSG-associated elevation of PD-L1 prompted us to next investigate the efficacy of immune checkpoint blockade in lean, obese, and formerly obese mice that lost weight by VSG or weight matched controls. While obese mice were resistant to immune checkpoint blockade, anti-PD-L1 potently impaired tumor progression after VSG through improved anti-tumor immunity. Thus, in formerly obese mice, surgical weight loss followed by immunotherapy reduced breast cancer burden.

cancer biology↗

bioRxiv2022

Protein Kinase C Delta Regulates Mononuclear Phagocytes and Hinders Response to Immunotherapy in Cancer

Checkpoint immunotherapy unleashes T cell antitumor potential which has revolutionized cancer treatment showing unprecedented long-term responses. However, most patients do not respond to immunotherapy which often correlates with a dysfunctional or immunosuppressive myeloid compartment. The mononuclear phagocyte system (MPS) is a sub-class of myeloid cells comprising monocytes, macrophages and dendritic cells which plays a crucial role in tissue homeostasis. However, accumulating evidence suggests that mononuclear phagocytes contribute to all phases of tumorigenesis including orchestrating inflammatory events during de novo carcinogenesis, contribution to the progression of established tumors and promotion of resistance to checkpoint blockade. Thus, targeting the MPS could be an effective strategy to enhance checkpoint blockade efficacy and promote control of tumors. Here, we found that protein kinase C delta (PKC{delta}), a serine/threonine kinase, is abundantly expressed by mononuclear phagocytes in several human and mouse tumors. PKC{delta}-/- mice were more resistant to growth of various cancers compared to wild-type mice and were more responsive to anti-PD-1 immunotherapy. Furthermore, we found that tumors from PKC{delta}-/- mice harbor a Th-1-skewed immune landscape including increased antigen cross-presentation and T cell activation. Depletion of mononuclear phagocytes in vivo altered tumor growth in wild-type mice, but not in PKC{delta}-/- mice. In addition, coinjection of PKC{delta}-/--deficient M2-like macrophages with cancer cells into wild-type mice markedly delayed tumor growth and significantly increased intratumoral T cell activation compared to wild-type M2-like macrophages coinjected with cancer cells. Finally, intrinsic loss of PKC{delta}-/- functionally reprogrammed macrophages and dendritic cells by promoting their antigen presenting and cross-presenting capacity and triggered type I and type II interferon signaling. Thus, PKC{delta} might be targeted to reprogram mononuclear phagocytes and augment checkpoint blockade efficacy.

immunology↗

bioRxiv2021

Comparative Multiomic Analysis Reveals Low T Cell Infiltration as the Primary Feature of Tobacco Use in HPV(+) Oropharyngeal Cancer

PurposeTobacco use is an independent adverse prognostic feature in human papillomavirus (HPV)-associated oropharyngeal squamous cell carcinoma (OPSCC). Despite this, the biologic features associated with tobacco use have not been systematically investigated in this population. We sought to characterize the genomic and immunologic features of HPV(+) OPSCC associated with tobacco use and adverse oncologic outcomes. Experimental DesignWhole exome sequencing of 47 primary HPV(+) OPSCC tumors was performed to investigate mutational differences associated with tobacco exposure. To characterize the tumor immune microenvironment (TIME), targeted mRNA hybridization was performed and immunohistochemical (IHC) staining was used to validate these findings. ResultsLow expression of transcripts in a T cell-inflamed gene expression profile (TGEP) was associated with tobacco use at the time of diagnosis and lower overall and disease-free survival. Tobacco use was associated with an increased proportion of T>C substitutions and a lower proportion of mutational signatures typically observed in HPV(+) OPSCC tumors, but was not associated with increases in mutational burden or the rate of recurrent oncogenic mutations. ConclusionsIn HPV(+) OPSCC, low T cell infiltration of primary tumors is associated with current tobacco use and worse oncologic outcomes. Rather than an increased mutational burden, tobaccos primary and clinically relevant association is immunosuppression of the primary TIME. An objective clinical assay like the TGEP, which quantifies immune infiltration of the primary TIME, may have value for HPV(+) OPSCC risk stratification in future clinical trials.

cancer biology↗

bioRxiv2021

Systematic analysis of SARS-CoV-2 infection of an ACE2-negative human airway cell

Established in vitro models for SARS-CoV-2 infection are limited and include cell lines of non-human origin and those engineered to overexpress ACE2, the cognate host cell receptor. We identified human H522 lung adenocarcinoma cells as naturally permissive to SARS-CoV-2 infection despite complete absence of ACE2. Infection of H522 cells required the SARS-CoV-2 spike protein, though in contrast to ACE2-dependent models, spike alone was not sufficient for H522 infection. Temporally resolved transcriptomic and proteomic profiling revealed alterations in cell cycle and the antiviral host cell response, including MDA5-dependent activation of type-I interferon signaling. Focused chemical screens point to important roles for clathrin-mediated endocytosis and endosomal cathepsins in SARS-CoV-2 infection of H522 cells. These findings imply the utilization of an alternative SARS-CoV-2 host cell receptor which may impact tropism of SARS-CoV-2 and consequently human disease pathogenesis.

microbiology↗

bioRxiv2020

CRTC1-MAML2 Establishes a PGC1α-IGF1 Circuit that Confers Vulnerability to PPARγ Inhibition

Mucoepidermoid carcinoma (MEC) is a life-threatening salivary gland cancer that is driven primarily by the transcriptional co-activator fusion CRTC1-MAML2. The mechanisms by which the chimeric CRTC1-MAML2 oncoprotein rewires gene expression programs that promote tumorigenesis remain poorly understood. Here, we show that CRTC1-MAML2 induces transcriptional activation of the non-canonical PGC-1 splice variant PGC-14, which regulates PPAR{gamma}-dependent IGF-1 expression. This mitogenic transcriptional circuitry is consistent across cell lines and primary tumors. CRTC1-MAML2 positive tumors are dominated by IGF-1 pathway activation and small molecule drug screens reveal that tumor cells harboring the fusion gene are selectively sensitive to IGF-1R inhibition. Furthermore, this dependence on autocrine regulation of IGF-1 transcription renders MEC cells susceptible to PPAR{gamma}inhibition with inverse agonists. These results yield insights into the aberrant co-regulatory functions of CRTC1-MAML2 and identify a specific vulnerability that can be exploited for precision therapy.

cancer biology↗